The Canadian Centre for Cyber Security and the U.S. Cybersecurity and Infrastructure Security Agency (CISA) issued Java4Shell patch alerts on November 17, 2026. The warnings followed the release of fixes for the newly disclosed Java runtime vulnerability.
Tracked as CVE-2026-114938, Java4Shell is a memory-corruption flaw in the hardware-accelerated Base64 decoder used by OpenJDK’s HotSpot Java Virtual Machine (JVM). Base64 represents binary data as text. Successful exploitation could disclose sensitive information, crash a Java process or allow remote code execution, according to the Cyber Centre.
Microsoft identifies affected, unpatched builds on the Java 17 line starting with 17.0.6, along with Java 21, 25 and 27. Exploitation requires attacker-controlled input to reach the vulnerable decoder on x86-64 hardware supporting the AVX-512 VBMI instruction set. An affected Java version alone doesn’t establish that an application is exploitable.
Microsoft says ARM64 systems and x86-64 processors without AVX-512 VBMI support aren’t known to be affected. Red Hat says OpenJDK 8 and 11 don’t contain the vulnerable code path. Oracle separately lists its Java 8 Enterprise Performance Pack build 8u511b33-perf among affected products, so the OpenJDK 8 exclusion shouldn’t be extended to every Java 8 distribution.
Oracle assigns the vulnerability a Common Vulnerability Scoring System (CVSS) 3.1 score of 8.1, rated high. Calif, the security firm that disclosed Java4Shell, assigns it 9.2, rated critical, under CVSS 4.0.
Enterprise products have different attack paths
Calif reported unauthenticated remote code execution in its testing of default configurations of applications including Keycloak, Atlassian Jira, Atlassian Confluence and Jenkins on affected systems. Keycloak’s team says some Base64 decoding occurs before authentication and warns against assuming a login requirement protects an affected instance.
Elastic describes a different boundary for Elasticsearch. Its advisory says versions 7.17.3 and later are vulnerable on affected CPU architectures, but code execution requires an authenticated user with at least read access. Unauthenticated attackers can cause denial of service against affected Elasticsearch nodes, according to the company.
Bundled Java can escape software inventories
CISA warns that Java runtimes may be included inside third-party applications, container images and appliances without appearing as separate software inventory entries. Its alert calls for identifying those installations alongside standalone Java deployments.
Keycloak’s November 18 advisory provides a specific example: the x86-64 container images tagged 26.8.2-1 and 26.8.2-2 bundled an affected OpenJDK 21.0.13 build. The team cautioned that image tags alone don’t establish exploitability because CPU features and effective JVM settings also matter.
Taken together, the advisories indicate that updating a host’s default Java installation isn’t sufficient evidence that an application’s separately bundled runtime has been fixed.
Fixed builds are available, but packaging matters
The Cyber Centre lists the following fixed Oracle Java SE builds in its alert.
| Oracle Java SE product | Fixed build listed |
|---|---|
| Java SE 17 | 17.0.21.1 |
| Java SE 21 | 21.0.13.1 |
| Java SE 25 | 25.0.5.1 |
| Java SE 27 | 27.0.1.1 |
| Java SE 8 Enterprise Performance Pack | 8u513-perf |
Those are Oracle build identifiers, not universal package requirements for every Java distribution or bundled product. Keycloak advises using a runtime or image whose vendor confirms that CVE-2026-114938 is fixed, rather than relying on the application version number alone.
Jenkins’ November 18 guidance calls for upgrading Java and restarting Jenkins to load the fix. Red Hat also says customer-built containers should be rebuilt when updated base images become available, extending remediation beyond replacing a package on the host.
Temporary controls do not replace patched runtimes
Jenkins and Keycloak both recommend the JVM startup flag -XX:-UseBASE64Intrinsics as an interim mitigation when an immediate runtime upgrade isn’t possible. The flag disables the vulnerable accelerated decoding path and requires restarting affected instances. Keycloak also calls for checking that the running JVM actually reports the optimization as disabled.
Microsoft has published Azure Web Application Firewall mitigations designed to detect and block known HTTP-based exploitation patterns. The company explicitly describes those measures as supplementary protection while security updates are applied, not a replacement for patching.
Research demonstrates exploitability, not attack scale
VulnCheck said its early reproduction found that triggering crashes was substantially easier than achieving remote code execution when attacker-controlled input reached the vulnerable decoder. Its code-execution assessment involved additional, target-specific work, including a memory-disclosure path to overcome address randomization.
In a November 18 report, SANS Internet Storm Center researcher Johannes Ullrich said one of its Java-specific honeypots was receiving many attacks, but none of the attacks observed there exploited Java4Shell. That observation was limited to the honeypot’s visibility and didn’t establish whether exploitation was occurring elsewhere.
CISA prioritizes internet-facing systems for patching. Its monitoring guidance also includes unexpected JVM crashes, rogue runtimes and suspicious Java processes as activity warranting attention.

Tech Help Canada Staff researches, writes, and reviews practical content for business owners and professionals. Our coverage spans business, marketing, SEO, technology, and the tools and systems people use to grow and operate online. We focus on clear, useful information backed by research, hands-on experience, and editorial review. Learn more about our team and editorial standards. Need help with something? Contact Us







